The field of predictive analytics is not without its own set of challenges
Like oil, data can be worthless in its original form.
By refining and distributing data and information, businesses can predict future trends and events — thus gaining commercial value from it.
Like most developments in the data economy, the field of predictive analytics is not without its own set of challenges. If you’re considering investing into scaling an algorithm to help your business across its digital strategy, here’s what you need to consider:
1. Privacy and ownership of data
Since the birth of the data economy, conflicts have raged on differing viewpoints as to which group has created the biggest value. The stakeholders are as follows:
- The producers and aggregators of data,
- The users who are represented in the data, and
- The consumers of data.
Notable commentary simply states that the tradeoff for using a service for free is that users are will to surrender their data to the service provider. It has also been argued that users should profit from the exploitation of their data. Policy experts have questioned whether the user has the right to access the data if it is anonymised.
The question is: Are you for a data monopoly, or is that anticapitalist? Twitter gradually placed restrictions on its data feeds, going against an open data dominant strategy. In doing so, its approach views data (and its exclusivity) as a profit centre.
Technical solutions can allow organizations to remove the metadata component that identifies an individual person, leaving only their footprints in the digital sand. When deciding your company’s stance on the issue, take into account the brand promise, customer perception and legalities of the matter, so an off-hand discovery of your data-use is not met with shock.
2. Analysis of user data
Sentiment and intent analysis represent the cornerstone of predictive analysis for advertisers. This is why search advertising is far more effective than display advertising in every way. Like with Google’s Waze, another application is how the system can predict of traffic patterns in real-time, thus allowing marketers to determine how best to serve location-specific offers that generate drive-through or physical footfall.
Technical solutions in this area rely on advanced middleware technology that can bring all this data from multiple sources. It then needs to be computed from an increasingly powerful machine learning community to make sense. The modelling steps (intent) and parameter fitting (sentiment) will be applied from a framework that considers generative models.
3. Data ecosystems and exchanges
It is often presumed and hoped that with the right systems in place, data can contextually service to exchanges and be used to reach the right audience at the right time in the right tone. So it’s fair to say that not all data holds equal value for all. For instance, Waze’s traffic prediction tools can help businesses map their supply chains and restaurants plan their time sensitive offers to drive footfall. But that data may not be valuable for a high end car dealership that is way past the affordability of the represented data set.
The development of pricing models for data exchanges and data ecosystems will vary on assumptions we make with regards to the value of data to each customer segment. When exchanges are not competing with potential customers on how the data is used, they represent a viable business model as data providers.
The views expressed here are of the author’s, and e27 may not necessarily subscribe to them. e27 invites members from Asia’s tech industry and startup community to share their honest opinions and expert knowledge with our readers. If you are interested in sharing your point of view, submit your post here.
Featured Image Copyright: wavebreakmediamicro / 123RF Stock Photo
The post Data driven: 3 predictive analytics challenges to watch out for in 2017 appeared first on e27.